<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-1007">
  <Title>Tense and Connective Constraints on the Expression of Causality</Title>
  <Section position="2" start_page="0" end_page="48" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The work reported in this paper aims at determining which constraints hold on the interaction between the expression of causality (with or without connective) and aspectual properties of the eventualities and of the tenses used to express them. As a matter of fact, it turns out that, at least in French, the choice of one of the two tenses pass4 simple (PS) or impar/ait (IMP) is not neutral with respect to the expression of causality, in particular realised by means of the connective &amp;quot;don c&amp;quot; (theref0re).</Paragraph>
    <Paragraph position="1"> It has been observed that even when concerned only with temporal localisation, it is not enough to characterize tenses if one do not take into account the effects of discourse relations between eventualities 1: (1a-b) (Molendijk, 1996);  Paul got fined. He was driving too fast c. La branche cassa.</Paragraph>
    <Paragraph position="2"> Paul tombait donc dans le vide The branch broke.</Paragraph>
    <Paragraph position="3"> Paul was therefore falling down d. Sa premiere demande fut refus6e.</Paragraph>
    <Paragraph position="4"> I1 en r6digeait donc une autre His first application was refused.</Paragraph>
    <Paragraph position="5"> He was therefore writing another one Our objective in this paper is twofold: we want to study systematically the interaction between the various parameters we have mentionned, in order to provide a general explanation for the acceptabilities that have been observed, and we also want these explanations be formulated in terms of &amp;quot;conditions of use&amp;quot;, so that our results are exploitable for text generation. As a matter of fact, the choice of an appropriate form to express a cause relation between events has proved a non trivial problem (Danlos, 1987; Danlos, 1998). Two parameters have been identified as playing an important role: first, the order of presentation (cause before consequence, or the contrary), and second, 2The contrast between PS and IMP is only roughly parallel to that between simple past and past progressive: e.g., the translation into French of a simple past can be either PS or IMP. We translate systematically IMP into past progressive, even when the glose does not have the same aspectuo-temporal properties as the French original. Similarly, &amp;quot;therefore&amp;quot; is only roughly equivalent to &amp;quot;done&amp;quot;.</Paragraph>
    <Paragraph position="6">  the presence (or absence) of a connective 3. The examples we deal with in this paper suggest that tenses, at least in French and in particular the choice between PS and IMP must also be taken into account.</Paragraph>
    <Paragraph position="7"> The assumptions we make for this work are the following.</Paragraph>
    <Paragraph position="8"> We assume the view on discourse adopted within the SRDT framework (Asher, 1993): in a coherent discourse, sentences are linked by discourse relations, which help finding anaphor antecedents, computing temporal localisations, etc. Here, we are concerned only with two discourse relations, both involving causality. We call the first one result, as in (Lascarides and Asher, 1993), it holds between two sentences when the main eventuality of the first one is the cause of the main eventuality of the second one. We assume here a very open notion of causality that we don't want to refine. 4 We call the other one explanation, it holds between two sentences when the cause is presented after its consequence, thus playing an explanation role for the first sentence. This configuration in interaction with &amp;quot;donc&amp;quot; has been studied in (Rossari and Jayez, 1997) where it is called &amp;quot;causal abduction&amp;quot;. null We adopt as a basis for the description of IMP the proposal made in the DRT framework (Kamp and Rohrer, 1983; Kamp and Reyle, 1993), amended with proposals made in French literature, in particular concerning the anaphoric properties of this tense (Tasmowski-De Ryck, 1985; Vet and Molendijk, 1985; Molendijk, 1994).</Paragraph>
    <Paragraph position="9"> At last, we adopt the description of the connective &amp;quot;donc&amp;quot; which is elaborated, in terms of conditions of use and semantic effects, in (Jayez and Rossari, 1998).</Paragraph>
    <Paragraph position="10"> We start by considering discourses where a cause is presented after its consequence (i.e., an explanation discourse relation should hold). We observe that a PS-IMP sequence is sufficient to achieve the explanation effect, but that this sequence is constrained by the type of causality 3(Danlos, 1988) shows the influence of many others parameters, like the voice active vs. passive, the presence of a relative clause, etc.</Paragraph>
    <Paragraph position="11"> 4For instance, we assume that causality holds between a branch breaking and John's falling (direct), but also between Jean's repairing his car and his driving it (indirect).</Paragraph>
    <Paragraph position="12"> at stake. We also notice that connectives do not seem to interfere with tenses in this case (SS 2). We then examine discourses where the cause is presented before the consequence. In the absence of connective, we observe that none of the acceptable forms automatically convey causality (SS 3.1). With the connective &amp;quot;donc&amp;quot;, causality is imposed by the connective, but in its turn it brings new constraints (SS 3.2). For each set of examples, we provide a general explanation and draw conclusions for text generation.</Paragraph>
  </Section>
class="xml-element"></Paper>